zep-memory

What problem does it solve?

Zep-memory helps AI agents retain important information across sessions by storing and retrieving it from a persistent knowledge graph, so future suggestions and actions respect prior context and discovered preferences.

Core Features & Use Cases

• Search-first retrieval for alignment: Finds relevant prior Preferences and Procedures before acting, reducing inconsistency.
• Immediate capture of reusable knowledge: Stores new requirements, facts, and step-by-step workflows as soon as they are discovered.
• Explicit group_id scoping for isolation: Prevents knowledge pollution by requiring intentional group_id management (per-user, per-project, per-worktree, etc.).
• Entity-focused, anti-bloat retrieval: Encourages entity_types filtering and targeted fact queries to avoid overly broad results.

Quick Start

Use zep-memory to store and later retrieve the user’s project-specific deploy procedure by searching for existing procedures first, then adding the new workflow as a Procedure under a group_id scoped to that project.